[PATCH 0/2] NFSv4/pNFS: fix client kernel panic from malformed GETDEVICEINFO

[PATCH 0/2] NFSv4/pNFS: fix client kernel panic from malformed GETDEVICEINFO

[Michael Bommarito]

A malicious or compromised NFSv4.1+ pNFS metadata server can panic any
pNFS-flexfile client by returning a GETDEVICEINFO body with a
multipath-DS count of >= 3 and exactly one valid (netid, uaddr) pair.
The unbounded inner loop in nfs4_ff_alloc_deviceid_node() (and the
parallel site in nfs4_fl_alloc_deviceid_node() for the legacy file
layout) keeps iterating after the first netaddr is decoded, consuming
the trailing version_count / version / minor words of the body as
opaque netid + uaddr pairs.  Both come out as zero-length strings;
xdr_stream_decode_string_dup() sets *str = NULL and returns 0; the
caller in nfs4_decode_mp_ds_addr() only checks "< 0" and immediately
calls strrchr(NULL, '.').

A QEMU/KASAN reproducer is described in the second patch.  The
shortest crashing GETDEVICEINFO body is 56 bytes, the panic is 5/5
deterministic at multipath_count = 10, and it fires before any
user-level read can complete on the first pNFS file the client
touches.

Patch 1 closes the NULL dereference itself by changing the two
xdr_stream_decode_string_dup() return-value checks in
nfs4_decode_mp_ds_addr() from "< 0" to "<= 0".  Patch 2 promotes
NFS4_PNFS_MAX_MULTI_CNT to include/linux/nfs4.h so flexfile and the
legacy file layout can share it, bounds the inner mp_count loop in
both drivers against that cap, and breaks the loop on the first NULL
return from nfs4_decode_mp_ds_addr() so a hostile server cannot drive
the decoder past a single malformed entry.  Either patch alone closes
the panic; both together close the latent unbounded-decode class.

The unbound on mp_count predates the flexfile driver: the same loop
exists in the legacy file layout since 35124a0994fc ("Cleanup XDR
parsing for LAYOUTGET, GETDEVICEINFO", 2011) and was carried into
flexfile by d67ae825a59d ("pnfs/flexfiles: Add the FlexFile Layout
Driver", 2014).  The NULL-deref site was introduced by 6b7f3cf96364
("nfs41: pull decode_ds_addr from file layout to generic pnfs") when
the netaddr decode was unified.  Stable backporting wanted for all
three.

Cc: stable@vger.kernel.org

Michael Bommarito (2):
  NFSv4/pNFS: reject zero-length r_addr in nfs4_decode_mp_ds_addr
  NFSv4/flexfile,filelayout: bound multipath DS count in GETDEVICEINFO

 fs/nfs/filelayout/filelayout.h            |  2 +-
 fs/nfs/filelayout/filelayoutdev.c         |  7 +++++--
 fs/nfs/flexfilelayout/flexfilelayoutdev.c | 10 ++++++++--
 fs/nfs/pnfs_nfs.c                         |  4 ++--
 include/linux/nfs4.h                      |  3 +++
 5 files changed, 19 insertions(+), 7 deletions(-)

--
2.47.3

[PATCH 1/2] NFSv4/pNFS: reject zero-length r_addr in nfs4_decode_mp_ds_addr

[Michael Bommarito]

nfs4_decode_mp_ds_addr() decodes the r_netid and r_addr opaques of a
netaddr4 from a GETDEVICEINFO multipath-DS body, then immediately
calls strrchr(buf, '.') to locate the port separator. Both decodes
use xdr_stream_decode_string_dup(), and the current code checks only
"nlen < 0" / "rlen < 0" before dereferencing the returned string.

When the on-wire opaque has length zero, xdr_stream_decode_opaque_inline()
returns 0 and xdr_stream_decode_string_dup() falls through to its
"*str = NULL; return ret" tail, leaving buf NULL with a return value
of 0. The "< 0" check does not catch this, and the next line is
strrchr(NULL, '.'), a kernel NULL pointer dereference reachable from
any pNFS-flexfile client mounted against a malicious or compromised
metadata server.

Reject the zero-length cases explicitly so the decoder fails with
-EBADMSG (treated as a malformed GETDEVICEINFO body) instead of
panicking the client.

Cc: stable@vger.kernel.org
Fixes: 6b7f3cf96364 ("nfs41: pull decode_ds_addr from file layout to generic pnfs")
Assisted-by: Claude:claude-opus-4-7
Signed-off-by: Michael Bommarito <michael.bommarito@gmail.com>
---
 fs/nfs/pnfs_nfs.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Reproduced via a malicious NFSv4.1/pNFS server returning a flexfile
GETDEVICEINFO body with multipath_count >= 3 and one valid
(netid, uaddr) pair.  Linux 7.0-rc7 + KASAN, QEMU/KVM.  Stock kernel:

  Oops: general protection fault [#1] SMP KASAN NOPTI
  KASAN: null-ptr-deref in range [0x0000000000000000-0x0000000000000007]
  RIP: 0010:strrchr+0x24/0x80
  Call Trace:
   nfs4_decode_mp_ds_addr+0xca/0x570
   nfs4_ff_alloc_deviceid_node+0x357/0x1370
   nfs4_find_get_deviceid+0x6b6/0xa90
   nfs4_ff_layout_prepare_ds+0x3cf/0xa40
   ff_layout_choose_ds_for_read+0x14c/0x350
   ff_layout_pg_init_read+0x2a2/0xb90
   ...
   nfs_file_splice_read+0xcf/0x190
   do_sendfile+0x8eb/0xdf0
  Kernel panic - not syncing: Fatal exception

Deterministic for any multipath_count >= 3; multipath_count <= 2
does not crash because the loop ends before consuming the malformed
trailing bytes.

Patched kernel rejects the same crafted body and a baseline
multipath_count = 1 mount + read completes normally.


diff --git a/fs/nfs/pnfs_nfs.c b/fs/nfs/pnfs_nfs.c
index 12632a706da88..0ff43dbcb7cd7 100644
--- a/fs/nfs/pnfs_nfs.c
+++ b/fs/nfs/pnfs_nfs.c
@@ -1075,14 +1075,14 @@ nfs4_decode_mp_ds_addr(struct net *net, struct xdr_stream *xdr, gfp_t gfp_flags)
 	/* r_netid */
 	nlen = xdr_stream_decode_string_dup(xdr, &netid, XDR_MAX_NETOBJ,
 					    gfp_flags);
-	if (unlikely(nlen < 0))
+	if (unlikely(nlen <= 0))
 		goto out_err;
 
 	/* r_addr: ip/ip6addr with port in dec octets - see RFC 5665 */
 	/* port is ".ABC.DEF", 8 chars max */
 	rlen = xdr_stream_decode_string_dup(xdr, &buf, INET6_ADDRSTRLEN +
 					    IPV6_SCOPE_ID_LEN + 8, gfp_flags);
-	if (unlikely(rlen < 0))
+	if (unlikely(rlen <= 0))
 		goto out_free_netid;
 
 	/* replace port '.' with '-' */
-- 
2.53.0

[PATCH 2/2] NFSv4/flexfile,filelayout: bound multipath DS count in GETDEVICEINFO

[Michael Bommarito]

Both the flexfile and the (legacy) file pNFS layout drivers decode a
multipath-DS count from a server-supplied GETDEVICEINFO body and then
iterate it via nfs4_decode_mp_ds_addr() without any upper bound. The
filelayout driver already caps the outer ds_num against
NFS4_PNFS_MAX_MULTI_CNT (== 256) but applies no equivalent cap to the
inner mp_count; the flexfile driver applies no cap on either.

In addition, both inner loops ignore a NULL return from
nfs4_decode_mp_ds_addr(), so once the on-wire data no longer matches
a valid netaddr4 encoding the loop is free to consume the trailing
bytes of the device_addr opaque as garbage netid + uaddr pairs. A
malicious or compromised pNFS metadata server can therefore drive
the inner loop indefinitely (up to 2^32 - 1 iterations) against a
fixed-size 56-byte body, with each iteration triggering an
allocation / kmemdup_nul cycle inside the decoder.

Promote NFS4_PNFS_MAX_MULTI_CNT from the filelayout private header to
include/linux/nfs4.h so both drivers (and any future pNFS layout
driver that decodes a multipath address list) bound the wire-level
field consistently. Apply the cap to the inner mp_count in both
drivers, matching the existing ds_num check, and bail on the first
NULL return so a server that lies about mp_count cannot quietly
extend the loop into the trailing layout-body bytes. This is
defense-in-depth on top of the companion patch which closes the
NULL-deref in nfs4_decode_mp_ds_addr(); either patch alone closes
the kernel-panic shape, both together close the latent
unbounded-decode class.

Cc: stable@vger.kernel.org
Fixes: 35124a0994fc ("Cleanup XDR parsing for LAYOUTGET, GETDEVICEINFO")
Fixes: d67ae825a59d ("pnfs/flexfiles: Add the FlexFile Layout Driver")
Assisted-by: Claude:claude-opus-4-7
Signed-off-by: Michael Bommarito <michael.bommarito@gmail.com>
---
 fs/nfs/filelayout/filelayout.h            |  2 +-
 fs/nfs/filelayout/filelayoutdev.c         |  7 +++++--
 fs/nfs/flexfilelayout/flexfilelayoutdev.c | 10 ++++++++--
 include/linux/nfs4.h                      |  3 +++
 4 files changed, 17 insertions(+), 5 deletions(-)

With this patch alone the crafted GETDEVICEINFO at multipath_count >= 3
is rejected at the bound check; malformed netaddr in the inner loop
bails on the first NULL return.  Either this patch or the companion
1/2 closes the panic; both together close the unbounded-decode class.

Baseline multipath_count = 1 mount + read completes normally.


diff --git a/fs/nfs/filelayout/filelayout.h b/fs/nfs/filelayout/filelayout.h
index c7bb5da93307d..03298f2e7cd69 100644
--- a/fs/nfs/filelayout/filelayout.h
+++ b/fs/nfs/filelayout/filelayout.h
@@ -39,7 +39,7 @@
  * RFC 5661 multipath_list4 structures.
  */
 #define NFS4_PNFS_MAX_STRIPE_CNT 4096
-#define NFS4_PNFS_MAX_MULTI_CNT  256 /* 256 fit into a u8 stripe_index */
+/* NFS4_PNFS_MAX_MULTI_CNT now in <linux/nfs4.h>; shared with flexfile. */
 
 enum stripetype4 {
 	STRIPE_SPARSE = 1,
diff --git a/fs/nfs/filelayout/filelayoutdev.c b/fs/nfs/filelayout/filelayoutdev.c
index 7226989ee4d53..c58c786dcf011 100644
--- a/fs/nfs/filelayout/filelayoutdev.c
+++ b/fs/nfs/filelayout/filelayoutdev.c
@@ -159,10 +159,13 @@ nfs4_fl_alloc_deviceid_node(struct nfs_server *server, struct pnfs_device *pdev,
 			goto out_err_free_deviceid;
 
 		mp_count = be32_to_cpup(p); /* multipath count */
+		if (mp_count > NFS4_PNFS_MAX_MULTI_CNT)
+			goto out_err_free_deviceid;
 		for (j = 0; j < mp_count; j++) {
 			da = nfs4_decode_mp_ds_addr(net, &stream, gfp_flags);
-			if (da)
-				list_add_tail(&da->da_node, &dsaddrs);
+			if (!da)
+				break;
+			list_add_tail(&da->da_node, &dsaddrs);
 		}
 		if (list_empty(&dsaddrs)) {
 			dprintk("%s: no suitable DS addresses found\n",
diff --git a/fs/nfs/flexfilelayout/flexfilelayoutdev.c b/fs/nfs/flexfilelayout/flexfilelayoutdev.c
index c40395ae08142..faed05cbe9f1c 100644
--- a/fs/nfs/flexfilelayout/flexfilelayoutdev.c
+++ b/fs/nfs/flexfilelayout/flexfilelayoutdev.c
@@ -78,12 +78,18 @@ nfs4_ff_alloc_deviceid_node(struct nfs_server *server, struct pnfs_device *pdev,
 		goto out_err_drain_dsaddrs;
 	mp_count = be32_to_cpup(p);
 	dprintk("%s: multipath ds count %d\n", __func__, mp_count);
+	if (mp_count > NFS4_PNFS_MAX_MULTI_CNT) {
+		dprintk("%s: multipath count %u greater than supported maximum %d\n",
+			__func__, mp_count, NFS4_PNFS_MAX_MULTI_CNT);
+		goto out_err_drain_dsaddrs;
+	}
 
 	for (i = 0; i < mp_count; i++) {
 		/* multipath ds */
 		da = nfs4_decode_mp_ds_addr(net, &stream, gfp_flags);
-		if (da)
-			list_add_tail(&da->da_node, &dsaddrs);
+		if (!da)
+			break;
+		list_add_tail(&da->da_node, &dsaddrs);
 	}
 	if (list_empty(&dsaddrs)) {
 		dprintk("%s: no suitable DS addresses found\n",
diff --git a/include/linux/nfs4.h b/include/linux/nfs4.h
index d87be1f25273a..bfc30baa8159a 100644
--- a/include/linux/nfs4.h
+++ b/include/linux/nfs4.h
@@ -767,6 +767,9 @@ enum pnfs_block_extent_state {
 	PNFS_BLOCK_NONE_DATA		= 3,
 };
 
+/* Maximum NFSv4.1 pNFS multipath data-server address count */
+#define NFS4_PNFS_MAX_MULTI_CNT		256
+
 /* on the wire size of a block layout extent */
 #define PNFS_BLOCK_EXTENT_SIZE \
 	(7 * sizeof(__be32) + NFS4_DEVICEID4_SIZE)
-- 
2.53.0

Re: [PATCH 2/2] NFSv4/flexfile,filelayout: bound multipath DS count in GETDEVICEINFO

[Anna Schumaker]

Hi Michael,

On Fri, May 22, 2026, at 9:40 PM, Michael Bommarito wrote:
> Both the flexfile and the (legacy) file pNFS layout drivers decode a
> multipath-DS count from a server-supplied GETDEVICEINFO body and then
> iterate it via nfs4_decode_mp_ds_addr() without any upper bound. The
> filelayout driver already caps the outer ds_num against
> NFS4_PNFS_MAX_MULTI_CNT (== 256) but applies no equivalent cap to the
> inner mp_count; the flexfile driver applies no cap on either.
>
> In addition, both inner loops ignore a NULL return from
> nfs4_decode_mp_ds_addr(), so once the on-wire data no longer matches
> a valid netaddr4 encoding the loop is free to consume the trailing
> bytes of the device_addr opaque as garbage netid + uaddr pairs. A
> malicious or compromised pNFS metadata server can therefore drive
> the inner loop indefinitely (up to 2^32 - 1 iterations) against a
> fixed-size 56-byte body, with each iteration triggering an
> allocation / kmemdup_nul cycle inside the decoder.
>
> Promote NFS4_PNFS_MAX_MULTI_CNT from the filelayout private header to
> include/linux/nfs4.h so both drivers (and any future pNFS layout
> driver that decodes a multipath address list) bound the wire-level
> field consistently. Apply the cap to the inner mp_count in both
> drivers, matching the existing ds_num check, and bail on the first
> NULL return so a server that lies about mp_count cannot quietly
> extend the loop into the trailing layout-body bytes. This is
> defense-in-depth on top of the companion patch which closes the
> NULL-deref in nfs4_decode_mp_ds_addr(); either patch alone closes
> the kernel-panic shape, both together close the latent
> unbounded-decode class.
>
> Cc: stable@vger.kernel.org
> Fixes: 35124a0994fc ("Cleanup XDR parsing for LAYOUTGET, GETDEVICEINFO")
> Fixes: d67ae825a59d ("pnfs/flexfiles: Add the FlexFile Layout Driver")
> Assisted-by: Claude:claude-opus-4-7
> Signed-off-by: Michael Bommarito <michael.bommarito@gmail.com>
> ---
>  fs/nfs/filelayout/filelayout.h            |  2 +-
>  fs/nfs/filelayout/filelayoutdev.c         |  7 +++++--
>  fs/nfs/flexfilelayout/flexfilelayoutdev.c | 10 ++++++++--
>  include/linux/nfs4.h                      |  3 +++
>  4 files changed, 17 insertions(+), 5 deletions(-)
>
> With this patch alone the crafted GETDEVICEINFO at multipath_count >= 3
> is rejected at the bound check; malformed netaddr in the inner loop
> bails on the first NULL return.  Either this patch or the companion
> 1/2 closes the panic; both together close the unbounded-decode class.
>
> Baseline multipath_count = 1 mount + read completes normally.
>
>
> diff --git a/fs/nfs/filelayout/filelayout.h b/fs/nfs/filelayout/filelayout.h
> index c7bb5da93307d..03298f2e7cd69 100644
> --- a/fs/nfs/filelayout/filelayout.h
> +++ b/fs/nfs/filelayout/filelayout.h
> @@ -39,7 +39,7 @@
>   * RFC 5661 multipath_list4 structures.
>   */
>  #define NFS4_PNFS_MAX_STRIPE_CNT 4096
> -#define NFS4_PNFS_MAX_MULTI_CNT  256 /* 256 fit into a u8 stripe_index */
> +/* NFS4_PNFS_MAX_MULTI_CNT now in <linux/nfs4.h>; shared with flexfile. */

I don't think we need the comment saying that the value has moved.

> 
>  enum stripetype4 {
>  	STRIPE_SPARSE = 1,
> diff --git a/fs/nfs/filelayout/filelayoutdev.c 
> b/fs/nfs/filelayout/filelayoutdev.c
> index 7226989ee4d53..c58c786dcf011 100644
> --- a/fs/nfs/filelayout/filelayoutdev.c
> +++ b/fs/nfs/filelayout/filelayoutdev.c
> @@ -159,10 +159,13 @@ nfs4_fl_alloc_deviceid_node(struct nfs_server 
> *server, struct pnfs_device *pdev,
>  			goto out_err_free_deviceid;
> 
>  		mp_count = be32_to_cpup(p); /* multipath count */
> +		if (mp_count > NFS4_PNFS_MAX_MULTI_CNT)
> +			goto out_err_free_deviceid;
>  		for (j = 0; j < mp_count; j++) {
>  			da = nfs4_decode_mp_ds_addr(net, &stream, gfp_flags);
> -			if (da)
> -				list_add_tail(&da->da_node, &dsaddrs);
> +			if (!da)
> +				break;
> +			list_add_tail(&da->da_node, &dsaddrs);
>  		}
>  		if (list_empty(&dsaddrs)) {
>  			dprintk("%s: no suitable DS addresses found\n",
> diff --git a/fs/nfs/flexfilelayout/flexfilelayoutdev.c 
> b/fs/nfs/flexfilelayout/flexfilelayoutdev.c
> index c40395ae08142..faed05cbe9f1c 100644
> --- a/fs/nfs/flexfilelayout/flexfilelayoutdev.c
> +++ b/fs/nfs/flexfilelayout/flexfilelayoutdev.c
> @@ -78,12 +78,18 @@ nfs4_ff_alloc_deviceid_node(struct nfs_server 
> *server, struct pnfs_device *pdev,
>  		goto out_err_drain_dsaddrs;
>  	mp_count = be32_to_cpup(p);
>  	dprintk("%s: multipath ds count %d\n", __func__, mp_count);
> +	if (mp_count > NFS4_PNFS_MAX_MULTI_CNT) {
> +		dprintk("%s: multipath count %u greater than supported maximum %d\n",
> +			__func__, mp_count, NFS4_PNFS_MAX_MULTI_CNT);
> +		goto out_err_drain_dsaddrs;
> +	}
> 
>  	for (i = 0; i < mp_count; i++) {
>  		/* multipath ds */
>  		da = nfs4_decode_mp_ds_addr(net, &stream, gfp_flags);
> -		if (da)
> -			list_add_tail(&da->da_node, &dsaddrs);
> +		if (!da)
> +			break;
> +		list_add_tail(&da->da_node, &dsaddrs);
>  	}
>  	if (list_empty(&dsaddrs)) {
>  		dprintk("%s: no suitable DS addresses found\n",
> diff --git a/include/linux/nfs4.h b/include/linux/nfs4.h
> index d87be1f25273a..bfc30baa8159a 100644
> --- a/include/linux/nfs4.h
> +++ b/include/linux/nfs4.h
> @@ -767,6 +767,9 @@ enum pnfs_block_extent_state {
>  	PNFS_BLOCK_NONE_DATA		= 3,
>  };
> 
> +/* Maximum NFSv4.1 pNFS multipath data-server address count */
> +#define NFS4_PNFS_MAX_MULTI_CNT		256

In the original location, this had a comment saying where the 256 came
from. Can you carry that over to here too, please?

Thanks,
Anna

> +
>  /* on the wire size of a block layout extent */
>  #define PNFS_BLOCK_EXTENT_SIZE \
>  	(7 * sizeof(__be32) + NFS4_DEVICEID4_SIZE)
> -- 
> 2.53.0